Dev

analysis/06_read_wwarn_k13.R

@@ -40,15 +40,25 @@ wwarn_k13 <- readxl::read_xls(here("analysis", "data-raw", "WWARN_K13_database_0
          PMID = as.numeric(pubMedId),
          notes = as.character(NA))
 
+# count studies
+length(unique(wwarn_k13$authors))
+
 # filter to Africa
 wwarn_k13 <- wwarn_k13 |>
   filter(continent == "Africa") |>
   arrange(country, site, year, mutation) |>
-  select(country, site, year, mutation, present, tested, lat, lon, PMID)
+  select(country, site, year, mutation, present, tested, lat, lon, PMID, authors)
+
+# count studies
+length(unique(wwarn_k13$authors))
 
 # drop studies with no PMID
 wwarn_k13 <- wwarn_k13 |>
-  filter(!is.na(PMID))
+  filter(!is.na(PMID)) |>
+  select(-authors)
+
+# count studies
+length(unique(wwarn_k13$PMID))
 
 # drop studies already in GEOFF
 s <- readRDS(here("analysis", "data-derived", "geoff_STAVE.rds"))
@@ -59,6 +69,9 @@ PMID_geoff <- s$get_studies() |>
 wwarn_k13 <- wwarn_k13|>
   filter(!(PMID %in% PMID_geoff))
 
+# count studies
+length(unique(wwarn_k13$PMID))
+
 # ------------------------------------------------------------------------
 # remove some studies and fix errors in others
 
@@ -93,10 +106,6 @@ wwarn_k13 <- wwarn_k13 |>
   filter(PMID != 33789667) |>
   bind_rows(readxl::read_excel(here("analysis", "data-raw", "PMID_k13_replace.xlsx"), sheet = "33789667"))
 
-# this study data has many errors, but replace at a later stage (see below)
-wwarn_k13 <- wwarn_k13 |>
-  filter(PMID != 33146722)
-
 # replace some bad characters in site names
 wwarn_k13 <- wwarn_k13 |>
   mutate(site = gsub("é", "e", site),
@@ -138,10 +147,16 @@ wwarn_k13 <- wwarn_k13 |>
               select(PMID, WT_variant),
             by = join_by(PMID))
 
+# count studies
+length(unique(wwarn_k13$PMID))
+
 # drop studies for which WT is NA, implying that none of the target loci are covered
 wwarn_k13 <- wwarn_k13 |>
   filter(!is.na(WT_variant))
 
+# count studies
+length(unique(wwarn_k13$PMID))
+
 # overlay variant on top of WT
 wwarn_k13 <- wwarn_k13 |>
   mutate(variant_string_combined = overlay_variant(variant_string, WT_variant))
@@ -188,7 +203,10 @@ wwarn_k13 <- wwarn_k13 |>
 
 # ------------------------------------------------------------------------
 
-# read back in Asua et al. data
+# replace Asua et al. data
+wwarn_k13 <- wwarn_k13 |>
+  filter(PMID != 33146722)
+
 df_asua <- readxl::read_excel(here("analysis", "data-raw", "PMID_k13_replace.xlsx"), sheet = "33146722")
 
 # merge with data on paper and tidy up to match wwarn_k13 format 
@@ -206,6 +224,27 @@ df_asua <- df_asua |>
 
 wwarn_k13 <- bind_rows(wwarn_k13, df_asua)
 
+# count studies
+length(unique(wwarn_k13$PMID))
+
+# ------------------------------------------------------------------------
+# import further cleaned datasets
+
+import_PMIDs <- c(28594879, 35144535, 33107096, 26483118, 31296223, 33864801, 33350925,
+                  27573632, 34216470, 32747827, 32822392, 34551228, 35477399, 31427297,
+                  31034031, 29458380, 32103124)
+
+wwarn_k13 <- wwarn_k13 |>
+  filter(!(PMID %in% import_PMIDs))
+for (i in seq_along(import_PMIDs)) {
+  wwarn_k13 <- wwarn_k13 |>
+    bind_rows(readxl::read_excel(here("analysis", "data-raw", "PMID_k13_replace.xlsx"),
+                                 sheet = as.character(import_PMIDs[i])))
+}
+
+# count studies
+length(unique(wwarn_k13$PMID))
+
 # ------------------------------------------------------------------------
 
 # save to file

analysis/07_read_wwarn_pd.R

@@ -43,6 +43,10 @@ wwarn_pd <- readxl::read_xls(here("analysis", "data-raw", "WWARN_partnerdrug_dat
   select(study_id, country, site, lon, lat, study_start, study_end, marker_type,
          tested, present, mixed_present, url)
 
+# count studies
+# have to go with url as I know some study_ids are duplicates of the same data
+length(unique(wwarn_pd$url))
+
 # filter to Africa countries
 wwarn_pd <- wwarn_pd |>
   filter(country %in% c("Angola", "Benin", "Burkina Faso", "Burundi", "Cameroon", "Central African Republic",
@@ -54,6 +58,9 @@ wwarn_pd <- wwarn_pd |>
                         "Zimbabwe", "Eritrea", "South sudan", "Libya", "Djibouti", "Algeria", "Botswana", 
                         "Somalia"))
 
+# count studies
+length(unique(wwarn_pd$url))
+
 # merge with data on PMID
 df_PMID <- read.csv(here("analysis", "data-raw", "WWARN_pd_PMID.csv"))
 wwarn_pd <- wwarn_pd |>
@@ -63,6 +70,9 @@ wwarn_pd <- wwarn_pd |>
 wwarn_pd <- wwarn_pd |>
   filter(!is.na(PMID))
 
+# count studies
+length(unique(wwarn_pd$PMID))
+
 # drop studies already in GEOFF
 s <- readRDS(here("analysis", "data-derived", "geoff_STAVE.rds"))
 PMID_geoff <- s$get_studies() |>
@@ -72,6 +82,9 @@ PMID_geoff <- s$get_studies() |>
 wwarn_pd <- wwarn_pd|>
   filter(!(PMID %in% PMID_geoff))
 
+# count studies
+length(unique(wwarn_pd$PMID))
+
 # remove specific PMIDs
 wwarn_pd <- wwarn_pd |>
   filter(PMID != 26236581) |> # data has some issues (numerator doesn't sum to denominator) and the original paper does not provide sufficient detail to correct
@@ -119,25 +132,32 @@ wwarn_pd <- wwarn_pd |>
   select(study_id, survey_id, country, site, lon, lat, study_start, study_end,
          variant_string, present, tested, PMID)
 
+# count studies
+length(unique(wwarn_pd$PMID))
+
 # replace some values manually from file where there are mistakes in the original
+# note that this includes 5 studies identified through cleaning of WWARN k13 data where PD data are also available. Hence the total studies will increase by 5.
 sheet_names <- excel_sheets(here("analysis", "data-raw", "PMID_pd_replace.xlsx"))
 for (i in seq_along(sheet_names)) {
   wwarn_pd <- wwarn_pd |>
     filter(PMID != as.numeric(sheet_names[i])) |>
     bind_rows(readxl::read_excel(here("analysis", "data-raw", "PMID_pd_replace.xlsx"), sheet = sheet_names[i]))
 }
 
+# count studies
+length(unique(wwarn_pd$PMID))
+
 # merge with info on paper
 df_paper <- read.csv(here("analysis", "data-raw", "paper_info.csv")) |>
   distinct(PMID, .keep_all = TRUE)
 wwarn_pd <- wwarn_pd |>
   left_join(df_paper,
             by = join_by(PMID))
 
-# ------------------------------------------------------------------------
-
-# tidy up and save to file
+# tidy up
 wwarn_pd <- wwarn_pd |>
   select(-c(url, doi))
 
+# save to file
 saveRDS(wwarn_pd, file = here("analysis", "data-derived", "wwarn_pd_clean.rds"))
+

analysis/09_read_who.R

@@ -38,6 +38,16 @@ who <- left_join(who_res, who_df, by = "ID") |>
   mutate(LATITUDE = as.numeric(LATITUDE),
          LONGITUDE = as.numeric(LONGITUDE))
 
+# Filter to k13, as mdr1 is copy number info only, and no crt left after previous filters
+who <- who |>
+  filter(MM_TYPE == "Pfkelch13")
+
+# count studies
+who |>
+  group_by(DATA_SOURCE, CITATION_URL) |>
+  summarise() |>
+  nrow()
+
 # Subset to African samples
 who_countries <- c("Angola", "Benin", "Botswana", "Burkina Faso", "Burundi", "Côte d’Ivoire", 
                    "Cabo Verde", "Cameroon", "Central African Republic", "Chad", 
@@ -52,9 +62,11 @@ who_countries <- c("Angola", "Benin", "Botswana", "Burkina Faso", "Burundi", "C
 who <- who |>
   filter(COUNTRY_NAME %in% who_countries)
 
-# Filter to k13, as mdr1 is copy number info only, and no crt left after previous filters
-who <- who |>
-  filter(MM_TYPE == "Pfkelch13")
+# count studies
+who |>
+  group_by(DATA_SOURCE, CITATION_URL) |>
+  summarise() |>
+  nrow()
 
 # filter to those with citation
 who <- who |>
@@ -67,6 +79,9 @@ who <- who |>
   rename(url = CITATION_URL) |>
   left_join(paper_info, by = join_by(url))
 
+# count studies
+length(unique(who$PMID))
+
 # drop studies already in GEOFF
 s <- readRDS(here("analysis", "data-derived", "geoff_STAVE.rds"))
 PMID_geoff <- s$get_studies()$PMID
@@ -75,13 +90,19 @@ PMID_geoff <- PMID_geoff[!is.na(PMID_geoff)]
 who <- who |>
   filter(!(PMID %in% PMID_geoff))
 
+# count studies
+length(unique(who$PMID))
+
 # drop studies already in WWARN
 s <- readRDS(here("analysis", "data-derived", "WWARN_STAVE.rds"))
 PMID_wwarn <- s$get_studies()$PMID
 
 who <- who |>
   filter(!(PMID %in% PMID_wwarn))
 
+# count studies
+length(unique(who$PMID))
+
 # get variant_num from proportion
 who <- who |>
   mutate(variant_num = round(as.numeric(PROPORTION) / 100 * SAMPLE_SIZE))
@@ -134,6 +155,9 @@ who <- who |>
 who <- who |>
   filter(!is.na(WT_variant))
 
+# count studies
+length(unique(who$PMID))
+
 # overlay variant on top of WT
 who <- who |>
   mutate(variant_string_combined = overlay_variant(variant_string, WT_variant))